Valencian researchers study the use of artificial intelligence to identify pneumonia in COVID-19

24/07/2020
Left to right, Alberto Albiol (iTEAM), Luis Caballero, Francisco Albiol (IFIC), Antonio Albiol (iTEAM)

 

A project of the IFIC (UV-CSIC), the UPV and FISABIO has been selected by the Carlos III Health Institute to study the application of Artificial Intelligence systems to the early detection of pneumonia in patients affected by the SARS-CoV-2 coronavirus, one of the most serious consequences of COVID-19 disease. The project aims to integrate these systems into clinical decision making.

Scientists from the Corpuscular Physics Institute (IFIC, University of Valencia-CSIC), from the Polytechnic University of Valencia (UPV) and from the Foundation for the Promotion of Health and Biomedical Research of the Valencian Community (FISABIO) have been selected in the COVID-19 special call of the Carlos III Health Institute, a reference body in Spain for Health research, to study the application of Artificial Intelligence systems that allow classifying pneumonia in patients affected by the SARS-CoV-2 coronavirus, one of the most serious consequences of the disease. The project aims to integrate these systems into clinical decision making.

Artificial Intelligence plays an important role in the future of Medicine, where the three P’s are required: prediction, prevention, and precision. Systems based on Artificial Intelligence help and improve the radiologist’s prediction.

Recently, the group of researchers from the Corpuscular Physics Institute and the Polytechnic University of Valencia participated in the development and implementation of machine learning techniques (one of the most used techniques in Artificial Intelligence) to help diagnose breast cancer using mammograms. This system was developed in an international competition, later clinically tested, and recently disseminated in a publication of high scientific impact.

In addition to achieving competitive results in breast cancer diagnostic models, the system developed by the Valencian researchers was a milestone in issues that affect the quality of diagnosis such as the use, exploitation, quality, and interpretation of data.

Now, the group of Valencian researchers apply this method to develop algorithms that classify the injuries caused by the coronavirus in the respiratory system through the analysis of medical images (x-rays or CT). This project has been selected by the Carlos III Health Institute, the Spanish centre of reference for health research, to receive funding within the special call to fight against COVID-19.

The first phases of the project have had the collaboration of researchers and experts in viral and imaging pathologies from FISABIO (Foundation for the Promotion of Health and Biomedical Research of the Valencian Community) and radiologists attached to the Biomechanics Institute of Valencia (IBV). The project arises from a collaboration between the researchers of the IFIC and the UPV with FISABIO in the DIRAC project (intelligent diagnosis for x-rays with implementation in integrated circuit), whose objective was the study of other pathologies and which was presented to the Valencian Agency of the Innovation (AVI).

“We are very pleased that this collaboration with the Polytechnic University of Valencia and FISABIO is recognised by the Carlos III Health Institute after several years dedicated to clinical work with machine learning systems”, says Francisco Albiol Colomer, researcher at the IFIC participating in the project with Luis Caballero Ontanava and Salvador Tortajada Velert.

The researchers intend to include a larger group of radiologists who facilitate the clinical interpretation of the algorithms, favour their use by companies and study the possibility of including images from other Autonomous Communities. This will allow the algorithm to be generalised so that it can be widely disseminated.

“In the project it is necessary to comply with the four fundamental axes of the FAIR principles (acronym for Findability, Accessibility, Interoperability and Reuse) of good practices for the management and administration of scientific data. For this reason, it is necessary to provide a good organisation of this data set (process known as data curation) to obtain good results in predictive models. It is in this field that our team, the FISABIO & CIPF Mixed Biomedical Imaging Unit, has great experience”, highlights Mariam de la Iglesia Vayá, FISABIO researcher participating in the project.

On behalf of the UPV, Antonio Albiol and Alberto Albiol, from the Institute of Telecommunications and Multimedia Applications (iTEAM) and Jon Ander Gómez and Roberto Paredes, from the Pattern Recognition and Human Language Technology (PRHLT) centre, are participating. Their work focuses on the development of different Deep Learning techniques for the automatic detection of COVID-19.

"The project represents an opportunity for us to put our experience in automatic recognition of medical images into a problem that requires an urgent solution and that in turn demands multidisciplinary collaboration," point out the UPV researchers.

The project is currently being carried out on the IFIC artificial intelligence computing platform, Artemisa, co-financed by the European Union through the Operational Program of the European Regional Development Fund (FEDER) of the Valencian Community 2014-2020 for the acquisition of R+D+i infrastructures and equipment.

 RX_COVID19_Tratada

Chest x-ray treated with AI to detect COVID-19. Credits: FISABIO, IFIC (UV-CSIC), UPV

 

More information:

http://webific.ific.uv.es/web/content/la-inteligencia-artificial-combinada-con-evaluaciones-de-radi%C3%B3logos-mejora-la-precisi%C3%B3n-de

 

https://www.isciii.es/Noticias/Noticias/Paginas/Noticias/FondoCOVID19MasProyectos37.aspx